Neuro-fuzzy based power system stabilizer of a multi-machine system

Full text




A project report submitted in partial fulfilment of the requirements for the award of the degree of

Master of Engineering (Electrical-Power)

Faculty of Electrical Engineering Universiti Teknologi Malaysia



Specially dedicated to my beloved parents for their endless support and encouragements.




I wish to express my sincere appreciation to my project supervisor, P.M. Md Shah Bin Majid for all his guidance, encouragement and support in completing the master project.

Most importantly, I would like to extend my gratitude to my parents for their support and patience during my pursuit for higher studies. They have encouraged me throughout my education, and I will always be grateful for their sacrifice, generosity and love.

Finally yet importantly, thanks to all the persons who directly or indirectly contributed because their perspective and guidance helped greatly to point me in the right direction until the completion of this thesis.




This thesis presents a study of neuro-fuzzy power system stabilizer (PSS) for stability enhancement of a multi-machine power system. In order to accomplish a stability enhancement, speed deviation (∆ω) and acceleration (∆ ) of the rotor synchronous generator were taken as the input to the neuro-fuzzy controller. These variables take significant effects on damping the generator shaft mechanical oscillations. The stabilizing signals were computed using the neuro-fuzzy membership function depending on these variables. Simulink Block Design and Matlab 7.6 were utilized in implementing the study. The simulations were tested under different types of conditions; the steady state operation, the three phases to ground fault, with load connected to the system, mechanical input power changes and reference voltage (Vref) step changes. The performance of the neuro-fuzzy power

system stabilizer was compared with the conventional power system stabilizer and without power system stabilizer.




Tesis ini membentangkan kajian Penstabil Sistem Kuasa neuro-fuzzy untuk meningkatkan tahap kestabilan suatu sistem penjana segerak. Untuk meningkatkan tahap kestabilan sistem ini, perbezaan kelajuan (∆ω) dan pecutan (∆ ) rotor pada penjana segerak digunakan sebagai isyarat masukan pengawal neuro-fuzzy. Isyarat masukan ini memberi kesan secara langsung bagi mengurangkan ayunan mekanikal terhadap aci penjana. Isyarat ini ditukarkan ke bentuk fungsi keanggotaan berdasarkan kelakuan penjana tersebut. Semua model dan analisis menggunakan perisian Simulink (MATLAB 7.6). Simulasi dijalankan berdasarkan beberapa keadaan iaitu keadaan mantap, kerosakan tiga fasa ke bumi, penjana berbeban, perubahan kuasa masukan mekanikal dan perubahan voltan rujukan (Vref).

Perbandingan dilakukan antara Penstabil Sistem Kuasa neuro-fuzzy dengan Penstabil Sistem Kuasa Konvensional dan sistem tanpa Penstabil Sistem Kuasa untuk mencari teknik kawalan yang terbaik.














LIST OF APPENDICES xv 1 INTRODUCTION 1 1.0 Introduction 1 1.1 Problem Statement 2 1.2 Objectives 3 1.3 Scopes 3 1.4 Thesis Organization 4


2.0 Introduction 5

2.1 Power System Stability 5

2.2 Power System Analysis 7


viii 2.2.2 Steady-state Stability Analysis of a Power


8 2.2.3 Transient Stability Analysis of a Power


8 2.2.4 Small Signal (Small Disturbances)

Stability 9

2.2.5 Transient Stability 9

2.3 Rotor Angle Stability 10

2.3.1 Synchronous Machine Characteristics 10 2.3.2 Power versus Angle Relationship 11

2.4 Low Frequency Oscillation 13

2.5 Power System Control 14

2.6 Power System Stabilizer 15

2.6.1 Conventional Lead-Lag Power System

Stabilizer 16 2.7 Conclusion 16 3 NEURO-FUZZY SYSTEM 17 3.0 Introduction 17 3.1 Definition 18 3.2 Types of Neuro-Fuzzy 19

3.2.1 Cooperative Neuro-Fuzzy System 19 3.2.2 Hybrid Neuro-Fuzzy System 19

3.3 ANFIS 20

3.4 Sugeno type fuzzy model 22

3.5 Back propagation Learning Algorithm 24

3.6 Neuro-Fuzzy Inference System 26

3.7 Neuro-Fuzzy 27

3.8 Conclusion 28


4.0 Introduction 29

4.1 Power System Model 29



4.3 ANFIS Implementation 32

4.4 Using ANFIS in Matlab 32

4.4.1 Structure and Parameter 32 4.4.2 Constraint of ANFIS 33

4.5 ANFIS Editor GUI 33

4.5.1 Training and Checking Data 34 4.6 ANFIS Design Procedure for PSS 35

4.7 Transient Response Analysis 36

4.8 Conclusion 36


5.0 Introduction 37

5.1 Simulation Studies 37

5.2 Steady State Operation 38 5.3 Three Phases to ground Fault 44

5.4 Generator Loading 48

5.5 Mechanical Input (Pm) Increase 52

5.6 Voltage Reference Step Changes 55


6.1 Conclusion 6.2 Future Work 60 61 REFERENCES 62 Appendices A-B 65-75





5.1 Speed deviation analysis at steady state operation 43 5.2 Terminal voltage analysis at steady state operation 43 5.3 Speed deviation analysis at three phases to ground



5.4 Terminal voltage analysis at three phases to ground fault


5.5 Speed deviation analysis at generator loading 50 5.6 Terminal voltage analysis at generator loading 51 5.7 Speed deviation analysis at mechanical input (Pm)



5.8 Terminal voltage analysis at mechanical input (Pm)



5.9 Speed deviation analysis at voltage reference step changes


5.10 Terminal voltage analysis at voltage reference step changes





2.1 The Power System Analysis 7

2.2 Power Transfer Characteristic of a two machine system


2.3 Diagram of a synchronous generator excitation control system


2.4 Block diagram of a typical Power System Stabilizer 16 3.1 A Hybrid Neuro-Fuzzy Controller 20

3.2 Structure of ANFIS 23

3.3 Structure of Neuro-Fuzzy Inference System 26

4.1 Power System Model 30

4.2 Simulink block diagram for neuro-fuzzy PSS 31

4.3 Checking data 34

4.4 Flow Chart of ANFIS Procedure 35

5.1 Subsystem for Machine 1 38

5.2 Subsystem for Machine 2 39

5.3 Subsystem for Area 1 with Machine 1 and 2 39 5.4 Subsystem for Area 2 with Machine 3 and 4 40 5.5 Two area, 4 machines and 11 bus system for Steady

State Operation


5.6 Speed deviation and terminal voltage response for steady state operation


5.7 Two area, 4 machines and 11 bus system for Three Phases to ground Fault


5.8 Speed deviation and terminal voltage response for three phases to ground fault


xii 5.9 Two area, 4 machines and 11 bus system for

Generator Loading


5.10 Speed deviation and terminal voltage response for generator loading


5.11 At area 1, Mechanical Input (Pm) Increase for

Machine 1


5.12 Speed deviation and terminal voltage response for mechanical input (Pm) increase


5.13 At area 1, Voltage Reference Step Changes for Machine 1


5.14 Speed deviation and terminal voltage response for voltage reference step changes




PSS - Power System Stabilizer

CPSS - Conventional Power System Stabilizer ANFIS - Adaptive Neuro-Fuzzy Inference System

FIS - Fuzzy Inference System

MSE - Means Square Error SSE - Sum of Square Error

J - Total moment of inertia of the rotor masses in kgm2 θm - Angular displacement of the rotor with respect to a

stationary axis in mechanical radians (rad)

Pe - Electrical power

Pm - Mechanical power

Ta - Net accelerating torque in Nm

Tm - Mechanical or shaft torque supplied by the prime mover

less retarding torque due to rotational losses in Nm Te - Net accelerating and electromagnetic torque in Nm

M - Inertia constant of machine (MJ/rad)

δ - Rotor angle perturbation

P - Active power (W)

ω - Rotor speed of synchronous machine (rad/s) ∆ω - Rotor speed deviation (rad/s)

TS - Synchronizing torque coefficient

TD - Damping torque coefficient

∆δ - Delta rotor angle

∆VT - Delta terminal voltage

f - Output frequency

N - Rotor speed in rpm


xiv PREF - Mechanical power references

PSV - Feedback through governor

TM - Turbine output torque

Vinf - Infinite bus voltage

VTREF - Terminal voltage reference

VT - Terminal voltage

VA - Voltage regulator output

VF - Field voltage

VE - Excitation system stabilizing signal

∆ω - Speed deviation

∆ - Acceleration

VPSS - PSS output signal

Q - Reactive power (VAR)

µ - Membership function

ci - The center of membership function





A Power System Model Parameter 65

A1 Synchronous Generator 65

A2 Excitation System 66

A3 Hydraulic Turbine and Generator 67 A4 Conventional Power System Stabilizer 68

A5 Transmission Line 69

A6 Three Phase Fault 70

B ANFIS Toolbox 71

B1 Training Data 71

B2 Data for training, checking and testing 72

B3 Generate FIS 72

B4 Feedforward Neural Network Architecture 73

B5 Training Error 73

B6 Multi Input Single Output 74

B7 5 membership function 74





1.0 Introduction

An interconnected power system basically consists of several essential components. They are namely the generating units, the transmission lines, the loads, the transformer, static VAR compensators and lastly HVDC lines [1]. During the operation of the generators, there may be some disturbances such as sustained oscillations in the speed or periodic variations in the torque that is applied to the generator. These disturbances may result in voltage or frequency fluctuation that may affect the other parts of the interconnected power system [1].

External factors, such as lightning, can also cause disturbances to the power system. All these disturbances are termed as faults. When a fault occurs, it causes the motor to lose synchronism if the natural frequency of oscillation coincides with the frequency of oscillation on the generators. With these factors in mind, the basic condition for a power system with stability is synchronism. Besides this condition, there are other important condition such as steady-state stability, transient stability, harmonics and disturbance, collapse of voltage and the loss of reactive power [2].

The stability of a system is defined as the tendency and ability of the power system to develop restoring forces equal to or greater than the disturbing forces to maintain the state of equilibrium [1].the system is also considered as a stable if it converges to another equilibrium position in the proximity of initial equilibrium point. If the physical state of the system differs such that certain physical variable increases with respect to time, the system is considered to be unstable. Therefore, the


2 system is said to remain stable when the forces tending to hold the machines in synchronism with one another are enough to overcome the disturbances [2]. The system stability that is of most concern is the characteristic and the behaviours of the power system after a disturbance.

Stability studies are generally categorized into two major areas: steady-state stability and transient stability [1]. Steady-state stability is the ability of the power system to regain synchronism after encountering small and slow disturbances. Example of small and slow disturbances is gradual power changes. Dynamic stability is the ability to regain synchronism after encountering small disturbances within a long time frame. The effects of large and sudden disturbances are known as transient stability. Examples of such faults are sudden outrage of a transmission line or the sudden addition of removal of the loads. The transient stability occurs when the power system is able to withstand the transient conditions following a major disturbance [2].

Power system stabilizers are used to generate supplementary control signals for the excitation system in order to damp the low frequency inter-area and intra-area oscillations [3]. The most widely used in existing power system is conventional power system stabilizer and has made a contribution in enhancing power system dynamic stability.

A supplementary excitation controllers referred to as power system stabilizer (PSS) have been added to synchronous generators to counteract the effect of high gain automatic voltage regulator (AVR) and other sources of negative damping [4]. The stabilizer should produce a component of electrical torque in order to provide a damping on the rotor which is in phase with speed variations.

1.1 Problem Statement

Power systems are complex nonlinear systems and often result in low-frequency power oscillations due to insufficient damping. Unfortunately, AVR provide negative damping that can make the system unstable, especially in large


3 system. Power system stabilizers provide this supplementary stabilizing signal and are widely used to suppress the generator electromechanical oscillations and enhance the overall stability of power systems [5].

Conventional PSS are based on linearized machine model and thus tuned at a certain operating point. Since, power systems are highly nonlinear complex systems, with configurations and parameters that change time, the conventional PSS design cannot guarantee its performance in a practical operating environment [5].

1.2 Objectives

The objectives of the project are:

i. To simulate and design a power system stabilizer (PSS) based on neuro-fuzzy

ii. To examine/ analyse the effect of a fault in a multi-machine power systems with PSS using MATLAB

iii. To make a performance comparison between without PSS, conventional PSS and with the neuro-fuzzy PSS

1.3 Scopes

The scope of this project will be undertaken in the following five development stages which are:

i. Study the power system stability phenomenon for multi-machine synchronous generator and control system using fuzzy logic and neural network techniques.

ii. Design a power system stabilizer based on neuro-fuzzy in MATLAB. iii. Simulate the controller design in MATLAB power system toolbox and


4 1.4 Thesis Organization

The rest of the thesis is organized as follow:

Chapter 2 describes the power system stability phenomena and power system stabilizers are used on generator. This chapter also discusses the basic theory of synchronous generator such as the model and related equations of the relationship between generator and load.

Chapter 3 discusses the basic of intelligent control techniques by neuro-fuzzy and some approaches of neuro-neuro-fuzzy system also will be explained. This chapter also briefly explains the structure and type of model used in neuro-fuzzy controller.

Chapter 4 discusses the parameter of power system modelling and proposed neuro-fuzzy PSS design. The power system modelling is illustrated by Simulink block diagram which presents highly interactive surroundings for simulation.

Chapter 5 presents the simulation results of the neuro-fuzzy design and discussion on the result from simulation using MATLAB. The structures of neuro-fuzzy are tested in several disturbances such as a three phases to ground fault, generator loading, mechanical input (Pm) increase and voltage reference step changes

in interconnected power system. The dynamic responses of synchronous machine during test conditions are performed and performances between without neuro-fuzzy and with neuro-fuzzy are discussed.

Chapter 6 gives the conclusion for the thesis and provide relevant recommendation to improve the stability and negative damping which include further research in this work.




[1] H. Saadat, Power System Analysis: McGraw-Hill International Edition, (1999).

[2] L. Z. Racz and B. Bokay, Power System Stability Amsterdam; New York: Elsevier, (1988).

[3] Larsen, E.V. and Swann, D.A., “Applying power system stabilizers, Part I, II, III”, IEEE Transaction on Power Apparatus and Systems, PAS-100 No. 6: Pp. 3017-3041, 1981.

[4] Othman Bin Jais, “Study of Fuzzy Logic Power System Stabilizer”, Thesis UTM, (2002).

[5] Ruhua You, Hassan J. Eghbali, and M. Hashem Nehrir, “An Online Adaptive Power system Stabilizer for Multimachine Systems”, IEEE Trans. On Power System, Vol 18, No 1: Pp 128-135, 2003.

[6] Kundur, P., Power System Stability and Control. United States of America: McGraw-Hill, (1994).

[7] Grainger, J.J., and Stevenson, W.D., Power System Analysis. New York: McGraw-Hill, (1994).

[8] Claudio A. Canizares and Fernaado L. Alvarado, “Point of Collapse and Continuous Methods for large AC/DC system”, IEEE Transaction on Power System, Volume 7, No 1: Pp 1-8, February 1993.


63 [9] Yi Guo, David J. Hill and Youyi Wang, “Nonlinear decentralized control of

largescale power systems”, Automatica, Pp 1275-1289, 2000.

[10] Juang, C.F.A. “TSK-Type Recurrent Fuzzy Network for Dynamics Systems Processing by Neural Network and Genethic Algorithms”,. IEEE Transaction on Fuzzy System, Vol 10, No 2: Pp 155-170, 2002.

[11] Aaron Francis Snyder, “Inter-Area Oscillation Damping with Power System Stabilizer and Synchronized Phasor Measurement”, Thesis Faculty of the Virgina Polytechnic Institute and State University, February 1997.

[12] N.L. Pahalawaththa, G.S. Hope and O.P. Malik, “An Implicit Self-Tuning Regulator as A Power System Stabilizer”, IEEE Transactions on Energy: Pp. 103-106, 1986.

[13] Elly Yuza Mahmud, “Fault Detection in Power Transmission Line using Neuro Fuzzy”, Thesis UTM, (2006).

[14] L.X. Wang and J.M. Mendel, “Fuzzy Basis Functions, Universal Approximation, and Orthogonal Least-Squares Learning”, IEEE Trans. On Neural Network, Vol 3, No 5: Pp 807-814, 1992.

[15] D. Nauck and R. Kruse, Designing Neuro Fuzzy Systems through Backpropagation, Paradigms and Practice: Pp 203-228,Kluwer, Boston, (1996).

[16] D. Nauck and R. Kruse, “Neuro-Fuzzy Classification with NEFCLASS”, in P. Kleinschmidt, A. Bachem, U. Derigs, D. Fischer, U. Leopold-Wildburger and R. Möhring (eds.), Operations Research Proceedings 1995: Pp. 294-299, 1996.

[17] J.S. Jang, “ANFIS: Adaptive-Network-Based Fuzzy Inference System”, IEEE Trans on Systems, Man and Cybernetics, Vol 23, No 3: Pp 665-684, 1993.


64 [18] T. Takagi and M. Sugeno, “Fuzzy Identifiction of Systems and it’s

Applications to Modeling and Control”, IEEE Transaction on Systems, Man and Cybernatics, Vol. 15: Pp116-132, 1983.

[19] Negnevitsky, M. Intelligent System and Soft Computing with Practical Application using Matlab. University of Tasmania, Australia. Unpublished. [20] Lin, C. T. ed. Neural Fuzzy Control System with Structure and Parameter

Learning. Singapore: World Scientific, (1994).

[21] W. Liu, G. K. V. Moorthy, and D. C.Wunsch, II, “Adaptive Neural Network based Power System Stabilizer Design”, IEEE Trans on Power System, Vol 4, No 5: Pp 2970–2975, 2003.

[22] D. K. Chaturvedi and O. P. Malik, “Neurofuzzy Power System Stabilizer”, IEEE Transaction on Energy Conversion, Vol 23, No 3: Pp 887-894, 2008. [23] Fuzzy Logic toolbox for use with MATLAB-user’s guide (version 2), The